The White House has decided it doesn’t need a new AI law. According to Briefs.co reporting from October 10, the administration is wagering that existing tort and product-liability doctrine, aimed at developers, can discipline the frontier labs without a fresh statutory regime. Treasury Secretary Scott Bessent told lawmakers last month that “The best way to guarantee safety is that the creators are liable for what they build and generate.” Former AI czar David Sacks has called the resulting exposure “the mother of all product liability lawsuits.”

That framing is politically tidy. It isn’t legally tidy. An agent’s behavior reflects both the developer that trained it and the business that deployed it, and no court has sorted that out. CNAS researcher Ben Hayum calls it “unsettled whether AI companies are legally responsible … or whether users share legal responsibility.”

Congress is already drafting around the ambiguity. The AI Agent Accountability Act’s criminal exposure, introduced October 1 by Senators Josh Hawley and Chris Murphy, would reach operators who knowingly deploy agents that recklessly cause hacking harm. Rep. Lori Trahan’s October 7 discussion draft extends the CLAIM Act’s deployer reach to any AI conduct that would be negligent, tortious, or criminal if done by a person. Neither is enacted.

Enforcement is moving faster than legislation. The Super Intelligence Force’s small-business footprint, led by DNI Jay Clayton, warned late Friday after the Anthropic breach that companies “must immediately disclose incidents” and remediate harm. The task force didn’t limit that to labs.

So a 12-person agency running an AI outreach tool is now a deployer, with no safe harbor and a developer-first liability narrative aimed over its head. Satya Nadella, writing on X Saturday, called for an “emergency brake” and human-readable logs of “every meaningful model action.” Plaintiffs’ lawyers heard him. Vendor indemnification stopped being boilerplate this week.

Sources

Sources